Ultrafast electron microscopy: Instrument response from the single-electron to high bunch-charge regimes

Accepted Manuscript Research paper Ultrafast Electron Microscopy: Instrument Response from the Single-Electron toHigh Bunch-Charge Regimes Dayne A. Plemmons, David J. Flannigan PII: DOI: Reference:

S0009-2614(17)30078-7 http://dx.doi.org/10.1016/j.cplett.2017.01.055 CPLETT 34493

To appear in:

Chemical Physics Letters

Received Date: Revised Date: Accepted Date:

2 January 2017 19 January 2017 23 January 2017

Please cite this article as: D.A. Plemmons, D.J. Flannigan, Ultrafast Electron Microscopy: Instrument Response from the Single-Electron toHigh Bunch-Charge Regimes, Chemical Physics Letters (2017), doi: http://dx.doi.org/ 10.1016/j.cplett.2017.01.055

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Ultrafast Electron Microscopy: Instrument Response from the Single-Electron to High Bunch-Charge Regimes Dayne A. Plemmons and David J. Flannigan* Department of Chemical Engineering and Materials Science, University of Minnesota, 421 Washington Avenue SE, Minneapolis, MN 55455, USA

Abstract: We determine the instrument response of an ultrafast electron microscope equipped with a conventional thermionic electron gun and absent modifications beyond the optical ports. Using flat, graphite-encircled LaB6 cathodes, we image space-charge effects as a function of photoelectron-packet population and find that an applied Wehnelt bias has a negligible effect on the threshold levels (>103 electrons per pulse) but does appear to suppress blurring at the upper limits (~105 electrons). Using plasma lensing, we determine the instrument-response time for 700-fs laser pulses and find that single-electron packets are laser limited (1 ps), while broadening occurs well below the space-charge limit.

*Author to whom correspondence should be addressed. Email: [email protected] Office: 612-625-3867 Fax: 612-626-7246

Page 1 of 28

1. Introduction The development of ultrafast electron diffraction and imaging methods has facilitated investigation of the structural origins of dynamic phenomena occurring across chemistry, physics, and materials science [1-6]. Central to these methods is the photoelectric generation of discrete electron packets from a metal or ceramic photocathode, typically with ultraviolet (UV) femtosecond (fs) laser pulses. Such discrete photoelectron packets act as the probe component of stroboscopic pump-probe experiments, the general concept of which forms the basis of fs alloptical studies of molecular dynamics pioneered by Zewail [7].

Under well-defined (but

instrument-specific) conditions [8], such packets can be used to resolve sub-picosecond, angstrom-scale motion.

Indeed, for properly equipped instruments, comprehensive

characterization of ultrafast dynamics can be achieved by accessing real-, reciprocal-, and energy-space structural and electronic information via imaging, diffraction, and spectroscopy, respectively [6,9-11]. Many studies and paradigm tests have been reported. These include elucidation of mechanisms of structural phase transitions [12-14]; determination of timescales for electron-phonon coupling [15,16]; and visualization of nanomechanical vibrational behavior [17,18], propagating acoustic phonons [19-23], and transient confined electromagnetic fields [24-27]. The use of discrete electron packets to probe transient structures introduces challenges not present in all-optical studies. Electron-electron repulsion occurring at the photoelectron source (space-charge) and during packet propagation from source to specimen (Coulombic repulsion) limits spatial and temporal resolutions via degradation of coherence lengths [8,28-31]. A variety of methods have been developed to overcome such limitations, including photoelectron-packet compression with synchronized electromagnetic fields [32-34] and

Page 2 of 28

operation at megavolt (as opposed to kilovolt) accelerating voltages [35,36]. These strategies have largely been implemented in dedicated electron-diffraction instruments, wherein dynamics are probed exclusively in reciprocal space and often with relatively large (100-μm diameter), parallel beams. An alternative method often employed in ultrafast electron microscopes – more specifically, conventional transmission electron microscopes (TEMs) modified for ultrafast stroboscopic operation – is to populate each probe packet with, on average, a single photoelectron, thus altogether circumventing the deleterious effects [37,38]. In this regime, one would expect the statistical temporal distribution of the probe component to approach the fs UV laser-pulse duration used to generate the photoelectrons [39-41]. Conducting ultrafast electron microscopy (UEM) experiments in the single-electron regime introduces additional practical challenges that limit the operational parameter space. Though fast-electron scattering cross-sections are large compared to X-ray photons, roughly a million or more are typically needed to generate sufficient signal-to-noise ratios. Thus, high laser repetition rates (e.g., MHz) and/or long acquisition times (e.g., tens of seconds or longer) must be used. Despite this, it has been shown that the UEM single-electron regime enables preservation of the intrinsic instrument spatial resolution (e.g., 2.3 Å in real-space images) [39,42,43].

Accordingly, the limiting factors in UEM experiments become relatively long

specimen relaxation times following fs excitation (which dictate the experiment repetition rate), as well as real-time environmental instabilities and specimen drift or other non-reversible motion occuring during acquisition (which ultimately limit real-space resolution). Note that such effects are expected to be significant in other UEM modalities as well, especially those employing beams focused onto the specimen (e.g., convergent-beam diffraction and scanning probe experiments). While the impact on parallel-beam diffraction is likely to be less significant, this

Page 3 of 28

modality suffers from challenges of data interpretation arising from signal averaging over large specimen areas, in addition to loss of phase information. Despite the challenges mentioned above, the UEM stroboscopic approach employing fs laser pulses can be used to study a wide variety of systems, materials, and phenomena in the combined nanometer-picosecond spatiotemporal parameter space (angstrom dimensions in reciprocal space). In this regard, UEM instruments equipped with relatively large, flat ceramic photocathodes (e.g., LaB6) can be used to generate photoelectron packets with populations spanning several orders of magnitude (i.e., from 1 to ~105 electrons per packet for fs UV laser pulses) [37,39,43]. Importantly, this enables operation at relatively low repetition rates when using packets with large populations, despite a commensurate reduction in temporal resolution from hundreds of fs to picoseconds [18,23]. While providing experimental versatility, detailed determination of the instrument characteristics (combined spatiotemporal resolution) is nontrivial owing to the complex interplay between photocathode position and properties (i.e., work function, geometry, Fermi distribution, etc.), electrostatic electron-gun biasing, and the magnitude of bunch charge [44]. Here, we characterize the instrument response of a UEM equipped with a standard Wehnelt assembly and absent any modifications beyond the optical periscopes used for stroboscopic operation. By using a plasma lensing effect, we are able to map the photoelectronpacket population and total temporal response as a function of fs laser-pulse energy. Following establishment of the threshold for space-charge saturation and the effects of Wehnelt biasing for fixed cathode positions, we determine the instrument-response time for a range of populations spanning from near the single-electron regime up to ~105 electrons per packet. We observe a power-law trend for the total instrument-response time as a function of photoelectron-packet

Page 4 of 28

population, with extrapolation to the single-electron regime revealing a laser-limited temporal resolution. For populations only slightly larger than a single electron, we experimentally observe a broadening of the response time beyond the laser limit. We discuss possible mechanisms of the observed behaviors, with emphasis on identifying and controlling potential sources of error and measurement variability.

2. Experimental Methods 2.1 UEM and laser setup Instrument-response properties were quantified for an FEI Tecnai Femto ultrafast electron microscope (Thermo Fisher Scientific) equipped with a self-biasing thermionic electron gun operated at 200 kV. The base (conventional) microscope is an FEI Tecnai G 2 20 TWIN TEM, which has been modified such that two optical periscopes are incorporated into the column, one at the electron gun and the other at the specimen goniometer [23]. For the studies reported here, 50- and 100-μm flat (diameter), truncated LaB6 photocathodes encircled with a concentric graphite ring (65- and 40-μm in annular width, respectively) were used (Applied Physics Technologies) [18]. It was found that the graphite ring improves beam stability and eliminates emission from the LaB6 shank. The 50- and 100-μm cathodes were positioned 250 and 350 μm, respectively, behind the Wehnelt aperture within the cylinder, which correspond to optimal values for maximizing collection efficiency into the illumination system of the Tecnai Femto [44]. For all experiments, a 2-mm diameter Wehnelt aperture and a custom 1.25 mm condenser aperture were used. Images were recorded with a Gatan Orius SC200B 4-megapixel (14 bit) CCD camera. The number of photoelectrons per packet (i.e., the photoelectron-packet population, ne-) was calculated by summing all counts detected with the CCD and dividing by the

Page 5 of 28

product of the phosphor ratio (four photons per electron), the acquisition time, and the laser repetition rate (200 kHz, unless otherwise noted). All experiments were conducted with a Yb:KGW diode-pumped solid-state fs laser (PHAROS, Light Conversion, 1.03-μm fundamental wavelength).

Photoelectrons were

generated by directing 257.5-nm (4.8-eV, fourth harmonic) pulses of 700-fs duration (full-width at half-maximum, FWHM) onto the LaB6 cathode through a vacuum-sealed sapphire optical window in the side of the UEM column and with an aluminum mirror within the column. We estimate the beam-width at the cathode tip to be ~100 μm (FWHM) for all experiments. Photoelectron-packet population was controlled by varying the UV pulse energy from 0.3 to 100 nJ. The specimen (next section) was excited with 1.03-μm pulses of 700-fs duration FWHM and a fluence of 10 mJ/cm2 (120-μm FWHM spot size measured ex-situ with a Newport LBP-1-USB beam profiler). Autocorrelation measurements of the 1.03-μm excitation-pulse duration were performed with an in-house-built frequency-resolved optical gating setup, while the 257.5-nm pulse durations were assumed to be the same as the near-IR owing to negligible dispersion relative to the pulse lengths used here.

The assumed duration of the UV pulse is further

discussed in Section 3 with respect to the measured UEM instrument-response time. For the laser used here, the pulses were stretched from 240 fs FWHM, as measured at 200 kHz. The time delay between the laser pump pulses and photoelectron probe packets was controlled with a 1-m motorized delay-stage (Aerotech PRO165LM; 2.5-μm accuracy and 750-nm bi-directional repeatability). 2.2 Time zero and the instrument-response measurement Temporal overlap of the pump laser pulse and probe photoelectron packet was determined via a plasma-lensing effect [19,45,46]. The specimen target used here was a 1000-

Page 6 of 28

mesh Cu TEM grid (Ted Pella). Each experiment consisted of a series of low-magnification bright-field UEM images of a single grid square acquired at different time delays (Figure 1a). Difference images generated from averaged pre-time-zero frames contain contrast arising only from the lensing effect. Analysis of the image series enables determination of time zero (defined as the time at which the image-intensity change reaches 50%) and the instrument-response time (σt), assuming plasma generation occurs on a timescale much shorter than the pump-pulse duration. Space-time contours generated by averaging along the Y-direction in the difference images at each time point illustrate the varied temporal response that occurs throughout the vacuum region within the grid square (Figure 1b).

To ensure sampling of only pseudo-

instantaneous dynamics, time traces having maximum spatial gradients in the difference images were isolated (vertical red line in Figure 1b; see Supplementary Figure 1). The extracted traces are normalized to the second percentile, baseline subtracted, and then fit with the function shown in Equation 1. (1) Here, A and t0 are the extracted amplitude and time-zero position, respectively. Note that σt is the standard deviation of a Gaussian peak function (i.e., the derivative of Equation 1).

3. Results and Discussion 3.1 Photoelectron-packet population from fs UV pulses In order to identify and characterize space-charge-limited effects, the photoelectronpacket population (ne-) as a function of UV laser-pulse energy was determined from UEM images of the cathode, with the applied Wehnelt bias effectively in an off or an on state. The electronic circuitry in the UEM used here is such that the Wehnelt bias and cathode heating Page 7 of 28

supply are coupled, as in a conventional TEM equipped with a thermionic gun. Here, a bias in the on state refers to cathode resistive heating (thus, an applied Wehnelt bias) to a point just below the threshold for detectable thermionic emission, while the off state refers to no resistive heating (no applied Wehnelt bias). Data for several applied Wehnelt biases between the defined off and on states are included in the Supplementary Content (see Supplementary Figure 2). Note that even in the off state, a gun crossover is generated and the cathode can be imaged, indicating a lensing effect still occurs at the Wehnelt aperture; optimal positioning of the LaB6 behind the aperture ensures a high collection efficiency into the anode [44]. Low and high UV pulse energies lead to qualitatively distinct photocathode images (Figure 2a). Structure is visible on the surface of the photocathode in the low-pulse-energy images but is absent at the high pulse energies, regardless of Wehnelt bias. By quantifying the behavior of ne- as a function of UV pulse energy (Figure 2b), one can see that the images were generated in the under-saturated and nearly-saturated (i.e., plateau) regimes for low and high pulse energies, respectively. This is expected, as increasing UV pulse energies with a fixed duration produce photoelectron packets of increasing initial density until space-charge effects begin to dominate the observed behavior. Less generally, a clear applied Wehnelt biasing effect can also be seen in the low-pulse-energy images in Figure 2a; relatively low image intensity appears at an xy-image position separate from the main beam. Such an effect could be due to over-focusing of photoelectrons emitted at off-normal angles. Note that this effect was not observed for the high pulse energies. Quantification of ne- as a function of UV pulse energy from 0.3 to 100 nJ for both cathodes shows a weak (perhaps negligible) dependence on applied Wehnelt bias for the fixed

Page 8 of 28

LaB6 positions used here (Figure 2b). The emission characteristics for each condition can be described with the image-charge-limited emission model outlined in Equation 2 [47].

(2)

Here, a quadratic dependence on pulse energy (F) is assumed in the pre-saturation limit, where linear (b1) and second-order (b2) coefficients represent one-photon and two-photon emission processes, respectively. Above the saturation limit, which occurs at a threshold energy F0 (nJ) and electron population

(electrons per pulse), logarithmic growth occurs as a result of

photoemission suppression due to the image-charge effect. Parameters for each of the curves shown in Figure 2b can be found in Table 1 of the Supplementary Content. For the 50-μm cathode,

is found to be 1.08E4 and 2.54E4, with F0 values of 33.8 nJ and 45.2 nJ, for the off

and on biasing states, respectively. significantly lower

Counter-intuitively, the 100-μm cathode exhibits

and F0 values of 4.90E3 and 16.7 nJ, and 3.97E3 and 12.6 nJ, for the off

and on states, respectively. Owing to the factor of four larger emission area, one would expect the 100-μm cathode to generate larger threshold values, all else being equal. Note, however, that a variety of factors affect the photoemission process and could contribute to the observed behavior, thus rendering direct comparison of different cathode sizes challenging. Such factors include cathode surface contamination, cathode surface topology, and the position and orientation of the cathode relative to the Wehnelt aperture [44]. Nevertheless, the

values (nearly 105 electrons per pulse)

found for the UEM instrument used here enable experiments to be conducted at relatively low repetition rates and moderate image magnifications [18,23], though a tradeoff in temporal resolution must be made with increasing ne- (discussed below). Page 9 of 28

3.2 Effect of biasing and packet population on space charge Insight into the relative extent of transverse space-charge spreading can be ascertained from the blurring that occurs in UEM cathode images. This effect is evident in the select frames presented in Figure 3a, where resolvable cathode features become blurred with increasing UV pulse energy. This blurring can be quantified by generating two-dimensional Fourier transforms of the cathode images; non-vanishing intensity at increasing spatial frequencies (μm-1) corresponds to sharper features in the real-space images. Further, the width of the radiallyaveraged, background-subtracted Fourier-space peak (here, defined as the width at which 95% of the intensity is contained in the peak, Δs95) is inversely proportional to the extent of real-space blurring (Figure 3b). In order to determine the effect of an applied Wehnelt bias (again, in the off and on states, as defined above) on space charge, the inverse of Δs95 was plotted against ne- (Figure 3c). Interestingly, dissimilar trends were observed for the off and on states; a power-law trend was observed for the off state, while logarithmic growth was seen for the on state, indicating application of a bias suppresses space-charge blurring at relatively large ne- for the cathode positions used here. Note that the optimal cathode position may shift from what is predicted for the off state upon application of the biasing level used here (i.e., just below the thermionicemission threshold) and cause a corresponding shift in crossover position in the cathode-anode gap, which could explain the observed effect. Importantly, it has been suggested that application of a bias could also be used to reverse increases in energy spread that occur during the initial propagation period, with a commensurate temporal lengthening of the photoelectron-packet distribution [39,43]. An evaluation of longitudinal broadening will be required to determine if such a tradeoff is at work in the UEM used here.

Page 10 of 28

3.3 Instrument response in the high bunch-charge regime Determination of the dependence of UEM instrument-response time on ne- for particular experimental conditions is required for isolation and quantification of initial ultrafast structural dynamics. As described in Section 2, this was accomplished here by using a plasma lensing effect in UEM imaging mode. Note that this can also be performed using electron energy-loss spectroscopy for suitably equipped UEM instruments [27,39,40,43].

Evaluation of the

instrument temporal response for photoelectron packets with varying populations is presented in Figure 4 (Wehnelt bias in the off state). Note that application of a Wehnelt bias showed no discernible difference from the off state for the cathode positions used here (see Supplementary Figure 3). In Figure 4a, a set of normalized kinetic image-intensity traces for measured values of ne- ranging from 30 to over 104 electrons per pulse are shown. The total instrument-response time (σt, standard deviation of the resulting Gaussian peak function) was found to increase with increasing ne-, from 1 ps at 30 electrons per pulse to 4 ps at 104 electrons per pulse. Further, it was found that the response exhibits a power-law trend (Figure 4a inset). In order to verify that the observed power-law trend results from effects of increasing bunch charge, the experiment was repeated three additional times, with each conducted on a different day (Figure 4b). The compiled triplicate data indeed follow a similar trend as shown in Figure 4a: a power-law exponent of 0.2 with an extrapolated single-electron-per-pulse temporal response of 400 fs (σt). Note that similar trends (in both functionality and magnitude) have been observed for comparable photoelectron-packet populations in ultrafast electron-diffraction instrumentation [8], as verified with a two-dimensional mean-field model depicting packet propagation in the cathode-anode gap. While this gap is significantly larger in TEMs modified for stroboscopic operation, the electric potential is similar (~10 MV/m). Thus, the similar

Page 11 of 28

temporal trend for different instruments suggests comparable overall longitudinal broadening behavior via the Boersch effect.

Interestingly, the extrapolated single-electron instrument-

response time of 422 fs (994 fs FWHM, Figure 4b) returns a duration of 705 fs FWHM for the photoelectron-packet distribution after deconvolution of the measured 700-fs FWHM IR pump pulse. This result indicates that the total instrument-response time is limited only by the laserpulse durations when operated in the single-electron regime, at least for the pulse widths used here. In addition to determining the instrument-response time, UEM imaging of the plasma lensing effect was used to find the position of time zero (here, the time at which the imageintensity change reaches 50%). Assuming sufficient plasma density is formed to cause onset of a detectable lensing effect on a timescale that is short relative to the total IR pump-pulse duration, the measured time zero can be approximated to represent true time zero (i.e., the precise moment of Gaussian-peak overlap of the pump pulse and probe packet distribution at the specimen). As noted in the caption, the kinetic image-intensity traces shown in Figure 4a have been arbitrarily aligned at their half-maximum for clarity. In actuality, however, the t0.5 time was found to vary (seemingly stochastically) over a range of approximately 6 ps for the three independent experiments shown in Figure 4b (see Supplementary Figure 4). Such behavior could arise from a variety of sources, such as variations in initial photoelectron-packet generation and propagation during acceleration (i.e., the cathode to anode gap) and minor instrument voltage fluctuations over the period of data acquisition (which could further limit temporal resolution). Relatedly, the measured time-zero position was observed to shift by 80 ps upon switching from the 50- to the 100-μm cathode; recall the 100-μm cathode was positioned 100 μm further behind the Wehnelt aperture than the 50-μm cathode. As with the total instrument response, precise knowledge of

Page 12 of 28

the position and behavior of time zero is critical for accurately isolating and quantifying initial ultrafast structural dynamics. 3.4 Instrument response approaching the single-electron limit While operating at large yet viable ne- values offers increased UEM experimental flexibility, the temporal resolution decreases; here, the increase is from a laser-limited boundary of 1 ps FWHM (extrapolated to one electron per pulse) to 10 ps FWHM for 10 4 electrons per pulse for 700-fs FWHM laser pulses (pump and probe; convoluted FWHM = 1 ps). Practically, image acquisition at relatively low instrument repetition rates is challenging in the singleelectron regime. A simple approach to overcoming this (as discussed in Section 1) is to increase the laser repetition rate at a fixed UV pulse energy – but only to the lowest value at which experiments can be reasonably conducted in order to minimize specimen heat accumulation. If this is accomplished via pulse picking, the laser-pulse durations will not change, and it may be expected that the instrument-response time will also remain unchanged, despite the increased average power being trained on the photocathode. Nevertheless, experimental rigor necessitates measurement of the response time when changing instrument conditions. Accordingly, in order to determine the effects of increased repetition rate for conditions more closely approaching the single-electron regime (compared to the studies described above), the instrument-response time was measured at 600 kHz and with a UV pulse energy such that newas measured to be 5.5 ± 1 electrons per pulse.

These settings allowed for reasonable

acquisition times (25 s) of images with sufficient signal-to-noise ratios for quantification. Figure 5a displays kinetic intensity traces generated from three separate UEM image scans conducted over a span of three hours in total. By fitting Equation 1 to data from all three runs, the

Page 13 of 28

instrument-response time (σt, standard deviation of the resulting Gaussian peak function) was found to be 540 ± 35 fs. As discussed above, the minimum possible UEM instrument-response time (absent any electron-packet compression or other countermeasures) for a given laser-pulse duration (σp) is , which is equivalent to σ of the laser autocorrelation function (σt,autocorr). Thus, σt,UEM will equal σt,autocorr in the laser-limited case.

Because dispersion is expected to be

negligible for the 700-fs pulses used here, it is assumed that the UV and IR pulse widths are the same and that the IR autocorrelation function is a reasonable approximation of the actual UV-IR convoluted response. As shown in Figure 5b, σt,autocorr for the IR pulses was measured to be 423 fs (996 fs FWHM). Recall that this value matches well with the extrapolated UEM singleelectron instrument-response time in Figure 3b (422 fs, 994 fs FWHM). Compared to this, σt,UEM for 5.5 electrons per pulse is broadened to 540 fs (1.3 ps FWHM). Thus, despite being near the single-electron regime (i.e., well outside the space-charge and Coulombic explosion regimes), an apparent broadening of the photoelectron-packet distribution was still observed. This may be explained by noting that the UV photon energy is significantly higher than the LaB 6 work function (4.8 eV compared to 2.7 eV). This could result in an instantaneous momentum spread of the photoemitted electrons. It is important to again emphasize that the UEM instrumentresponse times measured here are specific to the employed optical-excitation conditions; isolation of general physical phenomena from instrument-specific trends requires myriad additional systematic studies.

4. Conclusions

Page 14 of 28

In conclusion, we have systematically mapped the space-charge and temporal instrumentresponse parameter space as a function of photoelectron-packet population and applied bias for a UEM equipped with a conventional thermionic electron gun and absent any modifications beyond incorporation of optical periscopes. For IR pump and UV probe laser pulses of 700-fs FWHM durations, demonstration of instrument-response times ranging from 1 to 10 ps (FWHM), for laser-limited single-electron packets to those containing ~105 electrons, enables UEM experiments to be conducted at relatively low repetition rates, as previously demonstrated [18,23]. Because the solid-state specimen is not refreshed in fs stroboscopic UEM, as is done for ultrafast molecular-beam experiments, operation at low repetition rates provides a means to allow for sufficient heat dissipation to occur between individual pump-probe events, thus minimizing deleterious effects. Further, precise determination of the UEM response time and time-zero position, for varied laser and TEM settings, is critical for isolating and quantifying the initial intrinsic ultrafast structural dynamics. Moving forward, it will be important to establish the limits of UEM instrument-response time for shorter laser-pulse durations in the singleelectron regime, and to also determine the optimal conditions for combined Wehnelt biasing, photoelectron-packet population, and temporal resolution.

It is expected that the results

described here will provide a more quantifiable framework upon which to base further developments and optimizations for achieving angstrom-fs real-space UEM imaging.

Acknowledgments This work was supported partially by the National Science Foundation through the University of Minnesota MRSEC under Award Number DMR-1420013 and partially by the Arnold and Mabel Beckman Foundation through a Beckman Young Investigator Award. The

Page 15 of 28

funding sources had no involvement in the study design; in the collection, analysis, and interpretation of the data; in the writing of the report; and in the decision to submit the article for publication.

Figure 1.

Plasma lensing, time zero, and instrument-response time via bright-field UEM

imaging. (a) Low-magnification difference images of a 1000-mesh Cu grid square acquired before (upper panel, t < 0) and after (lower panel, t > 0) time zero. The color bar [shown to the right of the time trace in (b)] for the image false coloring shows cooler colors indicate electron depletion (i.e., a reduction in image intensity). The red box in the lower panel indicates the area over which the space-time contour plot in (b) was generated. The unchanged yellow border is the Cu grid, and the centered square (within which the intensity change occurs) is vacuum. (b, upper panel) Space-time contour plot generated by averaging over the Y image direction for each time delay. The red vertical line near X = 3.7 μm is the region from which the time trace was generated. (b, lower panel) Time trace illustrating the instrument-limited onset of electron depletion (i.e., the blue region in the difference images).

Figure 2. Photoemission characteristics as a function of UV pulse energy and applied Wehnelt bias. (a) Images of the 100-μm cathode acquired for photoelectron packets generated with UV pulse energies of 0.3 nJ (low pulse energy, upper-two panels) and 90 nJ (high pulse energy, lower-two panels) with the Wehnelt bias in the off and on states (left-two and right-two panels, respectively). Note that acquisition times for the low pulse-energy images were adjusted such that contrast levels were approximately the same as for the high pulse-energy images. (b) Photoelectron-packet population (ne-) versus UV pulse energy for the 50- and 100-μm cathodes

Page 16 of 28

(squares and circles, respectively) with the Wehnelt bias in the off and in the on states (red and black data points and model curves, respectively). The model and the counter-intuitive order-ofmagnitude reduction in ne- for the larger photocathode are discussed in the text.

Figure 3. Quantification of space-charge effects via UEM image analysis. (a) Select images of the 100-μm cathode generated with a range of UV pulse energies. The pulse energies (nJ) used to generate the select frames are shown in the lower-right corner of each panel. (b) Twodimensional discrete Fourier transforms of select images generated with measured ne- values of 24 (black), 440 (red), and 6160 electrons per pulse (blue). (c) Effective spatial coherence [(Δs95)1

, where Δs95 corresponds to the width at which 95% of the intensity is contained in the peak] as

a function of ne- with the bias in the off (red circles) and on (black squares) states. The solid red and black curves are power-law and logarithmic fits to the data, respectively, which are included to highlight the respective trends.

Figure 4. Instrument-response time (σt) as a function of ne-. (a) Normalized kinetic imageintensity traces generated from photoelectron-packet populations of 30 (blue triangles), 370 (light-blue inverted triangles), 2350 (red circles), and 10690 (black squares) electrons per pulse. The corresponding solid colored sigmoidal curves are fits of Equation 1 to the data. The intensity-change behavior is inverted compared to what is shown in Figure 1b for ease of fitting. In addition, the traces have been temporally aligned (for clarity) such that time zero occurs for each at the t0.5 intensity-change point. The inset shows σt as a function of ne- for the four measured values (black squares) and a power-law fit to the data (dashed red line).

Page 17 of 28

(b)

Compilation of three individual UEM time scans conducted on separate days (Run 1, 2, and 3) illustrating the behavior of σt with varying ne- and a power-law fit to the data (dashed red line).

Figure 5. Instrument-response time at an elevated repetition rate and near the single-electron regime. (a) Normalized kinetic intensity traces compiled from three separate image scans (Run 1, 2, and 3). The data is temporally shifted such that each run has approximately the same timezero position (t0 = 0 ± 82 fs). The solid-green sigmoid curve is the mean of the fits to each run. (b) Laser autocorrelation data (black squares; error bars represent one standard deviation) and UEM temporal instrument-response function (green peak function), which is the derivative of the Equation-1 (error function) fit to the kinetic traces. All σ values are the standard deviation of a Gaussian peak function.

References [1]

A.H. Zewail, Annu. Rev. Phys. Chem. 57 (2006) 65.

[2]

R.J.D. Miller, Annu. Rev. Phys. Chem. 65 (2014) 583.

[3]

F. Carbone, Eur. Phys. J.: Appl. Phys. 54 (2011) 33503.

[4]

M. Centurion, J. Phys. B 49 (2016) 062002.

[5]

W.E. King, G.H. Campbell, A. Frank, B. Reed, J.F. Schmerge, B.J. Siwick, B.C. Stuart, P.M. Weber, J. Appl. Phys. 97 (2005) 111101.

[6]

D.A. Plemmons, P.K. Suri, D.J. Flannigan, Chem. Mater. 27 (2015) 3178.

[7]

A.H. Zewail, J. Phys. Chem. A 104 (2000) 5660.

[8]

A. Gahlmann, S.T. Park, A.H. Zewail, Phys. Chem. Chem. Phys. 10 (2008) 2894.

[9]

A.H. Zewail, Science 328 (2010) 187.

[10]

D.J. Flannigan, A.H. Zewail, Acc. Chem. Res. 45 (2012) 1828.

Page 18 of 28

[11]

A. Adhikari, J.K. Eliason, J. Sun, R. Bose, D.J. Flannigan, O.F. Mohammed, ACS Appl. Mater. Interfaces 9 (2017) 3.

[12]

P. Baum, D.-S. Yang, A.H. Zewail, Science 318 (2007) 788.

[13]

N. Gedik, D.-S. Yang, G. Logvenov, I. Bozovic, A.H. Zewail, Science 316 (2007) 425.

[14]

K. Haupt, M. Eichberger, N. Erasmus, A. Rohwer, J. Demsar, K. Rossnagel, H. Schwoerer, Phys. Rev. Lett. 116 (2016) 016402.

[15]

Z. Tao, T.-R.T. Han, C.-Y. Ruan, Phys. Rev. B 87 (2013) 235124.

[16]

E.M. Mannebach, R. Li, K.A. Duerloo, C. Nyby, P. Zalden, T. Vecchione, F. Ernst, A.H. Reid, T. Chase, X. Shen, S. Weathersby, C. Hast, R. Hettel, R. Coffee, N. Hartmann, A.R. Fry, Y. Yu, L. Cao, T.F. Heinz, E.J. Reed, H.A. Dürr, X. Wang, A.M. Lindenberg, Nano Lett. 15 (2015) 6889.

[17]

O.-H. Kwon, B. Barwick, H.S. Park, J.S. Baskin, A.H. Zewail, Nano Lett. 8 (2008) 3557.

[18]

D.T. Valley, V.E. Ferry, D.J. Flannigan, Nano Lett. 16 (2016) 7302.

[19]

H. Park, X. Wang, S. Nie, R. Clinite, J. Cao, Phys. Rev. B 72 (2005) 100301.

[20]

A. Bugayev, A. Esmail, M. Abdel-Fattah, H.E. Elsayed-Ali, AIP Adv. 1 (2011) 012117.

[21]

A. Yurtsever, S. Schaefer, A.H. Zewail, Nano Lett. 12 (2012) 3772.

[22]

R.P. Chatelain, V.R. Morrison, B.L.M. Klarenaar, B.J. Siwick, Phys. Rev. Lett. 113 (2014) 235502.

[23]

D.R. Cremons, D.A. Plemmons, D.J. Flannigan, Nat. Commun. 7 (2016) 11230.

[24]

A. Feist, K.E. Echternkamp, J. Schauss, S.V. Yalunin, S. Schäfer, C. Ropers, Nature 521 (2015) 200.

[25]

L. Piazza, T.T. Lummen, E. Quinonez, Y. Murooka, B.W. Reed, B. Barwick, F. Carbone, Nat. Commun. 6 (2015) 6407.

[26]

A. Yurtsever, R.M. van der Veen, A.H. Zewail, Science 335 (2012) 59.

[27]

B. Barwick, D.J. Flannigan, A.H. Zewail, Nature 462 (2009) 902.

[28]

B.J. Siwick, J.R. Dwyer, R.E. Jordan, R.J.D. Miller, J. Appl. Phys. 92 (2002) 1643.

[29]

B.W. Reed, J. Appl. Phys. 100 (2006) 034916.

[30]

M.R. Armstrong, K. Boyden, N.D. Browning, G.H. Campbell, J.D. Colvin, W.J. DeHope, A.M. Frank, D.J. Gibson, F. Hartemann, J.S. Kim, W.E. King, T.B. LaGrange, B.J. Pyke,

Page 19 of 28

B.W. Reed, R.M. Shuttlesworth, B.C. Stuart, B.R. Torralva, Ultramicroscopy 107 (2007) 356. [31]

Z. Tao, H. Zhang, P.M. Duxbury, M. Berz, C.-Y. Ruan, J. Appl. Phys. 111 (2012) 044316.

[32]

R.P. Chatelain, V.R. Morrison, C. Godbout, B.J. Siwick, Appl. Phys. Lett. 101 (2012) 081901.

[33]

C. Kealhofer, W. Schneider, D. Ehberger, A. Ryabov, F. Krausz, P. Baum, Science 352 (2016) 429.

[34]

J. Maxson, D. Cesar, G. Calmasini, A. Ody, P. Musumeci, D. Alesini, arXiv:1611.01701 (2016).

[35]

S.P. Weathersby, G. Brown, M. Centurion, T.F. Chase, R. Coffee, J. Corbett, J.P. Eichner, J.C. Frisch, A.R. Fry, M. Guhr, N. Hartmann, C. Hast, R. Hettel, R.K. Jobe, E.N. Jongewaard, J.R. Lewandowski, R.K. Li, A.M. Lindenberg, I. Makasyuk, J.E. May, D. McCormick, M.N. Nguyen, A.H. Reid, X. Shen, K. Sokolowski-Tinten, T. Vecchione, S.L. Vetter, J. Wu, J. Yang, H.A. Dürr, X.J. Wang, Rev. Sci. Instrum. 86 (2015) 073702.

[36]

P. Zhu, Y. Zhu, Y. Hidaka, L. Wu, J. Cao, H. Berger, J. Geck, R. Kraus, S. Pjerov, Y. Shen, R.I. Tobey, J.P. Hill, X.J. Wang, New J. Phys. 17 (2015) 063004.

[37]

V.A. Lobastov, R. Srinivasan, A.H. Zewail, Proc. Natl. Acad. Sci. U.S.A. 102 (2005) 7069.

[38]

M. Aidelsburger, F.O. Kirchner, F. Krausz, P. Baum, Proc. Natl. Acad. Sci. U.S.A. 107 (2010) 19714.

[39]

L. Piazza, D.J. Masiel, T. LaGrange, B.W. Reed, B. Barwick, F. Carbone, Chem. Phys. 423 (2013) 79.

[40]

D.A. Plemmons, S.T. Park, A.H. Zewail, D.J. Flannigan, Ultramicroscopy 146 (2014) 97.

[41]

A. Feist, N. Bach, N. Rubiano da Silva, T. Danz, M. Möller, K.E. Priebe, T. Domröse, J.G. Gatzmann, S. Rost, J. Schauss, S. Strauch, R. Bormann, M. Sivis, S. Schäfer, C. Ropers, arXiv:1611.05022 (2016).

[42]

B. Barwick, H.S. Park, O.-H. Kwon, J.S. Baskin, A.H. Zewail, Science 322 (2008) 1227.

[43]

K. Bücker, M. Picher, O. Crégut, T. LaGrange, B.W. Reed, S.T. Park, D.J. Masiel, F. Banhart, Ultramicroscopy 171 (2016) 8.

[44]

E. Kieft, K.B. Schliep, P.K. Suri, D.J. Flannigan, Struct. Dyn. 2 (2015) 051101.

[45]

J.R. Dwyer, C.T. Hebeisen, R. Ernstorfer, M. Harb, V.B. Deyirmenjian, R.E. Jordan, R.J.D. Miller, Philos. Trans. R. Soc., A 364 (2006) 741. Page 20 of 28

[46]

C.M. Scoby, R.K. Li, P. Musumeci, Ultramicroscopy 127 (2013) 14.

[47]

P.L.E.M. Pasmans, D.C. van Vugt, J.P. van Lieshout, G.J.H. Brussaard, O.J. Luiten, Phys. Rev. Accel. Beams 19 (2016) 103403.

Page 21 of 28

Page 22 of 28

Page 23 of 28

Page 24 of 28

Page 25 of 28

Page 26 of 28

Highlights: 

Plasma lensing is used to determine the instrument response of a thermionic UEM



Space-charge threshold is determined for biasing and electron-packet population



Response time varies from 1 to 10 ps for 700-fs laser pulses and increasing bunch charge



Single-electron regime is found to be laser limited for 700-fs pulses

Page 27 of 28

Graphical abstract

Page 28 of 28